Classificação do estado de sonolência para motoristas com base em eletroencefalograma, redes neurais artificiais e algoritmos genéticos.

Abstract

Most of the time, the human brain is processing information to react according to the stim- uli received through the sensory nervous system. The mental fatigue is a state of drowsiness and low attention that increases the reaction time of an individual. In the traffic, the short reaction time is crucial to avoid accidents, what means that mental fatigue is a risk factor for drivers, especially when driving through highways in which higher speeds are normally al- lowed. This work describes a method to classify EEG signals in drowsiness by using artificial neural networks optimized by genetic algorithms. The input parameters were calculated for groups of 256 samples (1 second window) of channel AF7 of a Muse portable EEG headset. The application of the proposed method resulted in 13.42% of classification error (confu- sion), 14.49% of mean squared error, 85.21% of sensitivity and 87.95% of specificity, using data acquired during sessions in a driving simulator.